Data networks are generally constructed using a number of edge switches, also referred to as “edge modules”, that are connected to local data sources and data sinks. The edge modules are interconnected by a network core that switches traffic between edge modules, as required. The network core may be a distributed core that includes a plurality of geographically distributed core modules. The network core may also be an agile core that is reconfigured to satisfy fluctuating traffic demands.
Links interconnecting the edge modules and the core modules support communications paths for transferring data between edge modules. Each of the communications paths has a predetermined granularity defined as an integer multiple of a capacity unit. A capacity unit is the minimum capacity, in bits per second, that may be assigned to a traffic stream. Traditionally, the granularity of links interconnecting edge modules with core modules is uniform. Several examples of architectures for modern data networks with links of uniform granularity are described in Applicant's copending United States patent applications referenced below.
In Applicant's copending U.S. patent application Ser. No. 09/286,431, filed on Apr. 6, 1999, and entitled SELF-CONFIGURING DISTRIBUTED SWITCH, a network architecture is described in which high-capacity electronic edge modules are interconnected by an agile channel-switching optical core. A global controller selects paths through the channel switched core and reconfigures the paths in response to dynamic changes in data traffic loads. Reconfiguration timing between the edge modules and the channel switch core modules is performed to keep reconfiguration guard time minimized.
In Applicant's copending U.S. patent application Ser. No. 09/475,139, filed on Dec. 30, 1999, and entitled AGILE OPTICAL-CORE DISTRIBUTED PACKET SWITCH, an architecture for an optical-core network in which a core module controller selects paths through an associated core module and reconfigures the paths in response to dynamic changes in data traffic loads is described. Switching latency in the core modules is masked so that the source edge modules need not disrupt data transmission during core reconfigurations. A slight surplus capacity is provided in the core modules to facilitate core reconfigurations. The reconfiguration functions of the edge modules and the core modules are coordinated to keep reconfiguration guard time minimized. This yields a high capacity, load-adaptive, self-configuring switch that can be widely distributed to serve a large geographical area.
In Applicant's copending U.S. patent application Ser. No. 09/550,489, filed Apr. 17, 2000, and entitled HIGH CAPACITY WDM-TDM PACKET SWITCH, a network architecture is described in which electronic edge modules are interconnected by space switches that are operated in a time division multiplexed (TDM) mode. The use of TDM permits a channel (typically a wavelength in a WDM transport medium) to be split into several sub-channels. This increases the number of edge modules that can be directly reached without a requirement for tandem switching. The network comprises a distributed channel switching core, the core modules being respectively connected by a plurality of channels to a plurality of high-capacity packet switch edge modules. Each core module operates independently to schedule paths between edge modules, and reconfigures the paths in response to dynamic changes in data traffic loads reported by the edge modules. Reconfiguration timing between the packet switch edge modules and the channel switch core modules is performed to keep reconfiguration guard time minimized.
In Applicant's copending U.S. patent application Ser. No. 09/624,079, filed Jul. 24, 2000, and entitled MULTI-DIMENSIONAL LATTICE NETWORK, a multi-dimensional lattice network that scales to capacities of the order of a Yotta b/s (1024 bits per second) includes a plurality of sub-nets of edge module switches interconnected by an agile switching core. A sub-net may be constructed as an agile optical-core network as described in patent applications referenced above, or as an agile electronic-based core, as described in U.S. patent application Ser. No. 09/550,489 referenced above.
Each of the above-described network architectures was designed around network links that support paths of a uniform granularity. It is well known that the elasticity and granularity of a path can have a considerable impact on a network's efficiency and control complexity. It is also well known that coarse paths (paths in which the capacity unit is very large, e.g., 10 Gb/s) reduce the number of control variables and thus simplify network control functions. However, if traffic volume does not warrant coarse path granularity, the use of coarse paths may dramatically reduce network efficiency. On the other hand, paths having a fine granularity (paths in which the capacity unit is relatively small, e.g., 100 Kb/s) tend to increase network efficiency since more direct paths can be established from each edge module to other edge modules in the network. This reduces the mean number of hops between edge modules. However, control complexity is increased, especially if the core modules are agile core modules, as described in Applicant's copending patent applications referenced above.
It is generally believed that the diversity of data traffic in modern data networks is increasing. Transfer rate requirements in networks with diverse traffic loads may vary from a few Kb/s to a few Gb/s, or even up to a Tb/s. In addition, data traffic patterns typically show that the bulk of data traffic originating from any edge switch is often directed to only a few other edge switches, the balance of the data traffic being widely distributed to a very large number of other edge switches. While existing networks can be configured to service such distributions of data traffic, they are not adapted to provide such service efficiently. Consequently, there is a need for a “multi-grained” network that is capable of economically handling extremes in transfer rate and distribution of traffic without unmanageable control complexity.